Prebid is gaining popularity among publishers as an open-source header bidding solution that offers many advantages to publishers. The key factors behind its success include increasing ad revenue, improving transparency and control, optimising user experience, and providing a flexible and scalable solution that evolves with the constantly changing digital advertising landscape.
Prebid offers two technically different approaches to header bidding: client-side and server-side implementations.
Implementation methods and deployment
In a client-side implementation, the header bidding process takes place in the user's browser. The browser sends bid requests to multiple demand partners simultaneously. This implementation is usually deployed using the Prebid.js library.
In a server-side implementation, the process occurs on a server instead of the user's browser. This approach may require more complex infrastructure and integration management than the client-side solution. For server-side deployment, you can either choose an external service provider or build the solution using the Prebid framework. Collaborating with a partner can be a more cost-effective option. (You can find potential partners listed here).
Strengths and weaknesses
The client-side solution is easier to deploy and more flexible, but having too many demand partners can negatively affect page load speed and user experience. Additionally, the lack of control over data can raise privacy concerns.
The server-side solution offers faster performance and a better user experience, as well as more precise data management. However, weaker user identification and more complex maintenance can pose challenges. Nevertheless, it is suitable for publishers who want to improve their advertising results and user experience in the long term.
Server-side is also the only option for utilising Prebid in mobile applications, which has, in part, fueled the growing interest in server-side solutions.
Tests for more in-depth performance analysis
Both implementation methods have their strengths, and choosing between them can be challenging. We decided to conduct practical tests to obtain up-to-date information on the performance of both implementations, assessing various key indicators.
We carried out a multivariate test in collaboration with one of our publisher clients to analyse the performance of three different variations. The publisher's traffic was divided into three samples as follows:
- 25% of the inventory for Client-side Prebid
- 25% of the inventory for Server-side Prebid
- 50% of the inventory for parallel bidding competition (Client-side + Server-side)
We conducted the tests using Relevant Yield, an independent and comprehensive tool for publishers to streamline processes and enhance advertising productivity. The SSPs we tested were Adform, Equativ, Magnite, Pubmatic and Xandr. The tests were implemented in both desktop and mobile environments.
In the tests, we observed the following effects on key performance indicators:
eCPM - We noticed significant variations in eCPM values between different SSPs - two of the tested SSPs yielded considerably higher eCPM values on the server side in a desktop environment. In mobile devices, eCPM values were more consistent across different samples.
Bid rate - Particularly in the desktop environment, Magnite, Equativ, and Xandr participated less in auctions in the server-side sample compared to the client-side sample. On mobile, Pubmatic's bid rate was in a class of its own on the server-side.
Viewability - In the server-side sample, ad viewability was significantly better, especially in the desktop environment.
Total revenues - The server-side sample was slightly behind in terms of performance compared to the client-side sample. Parallel bidding competition was the most efficient method.
Summary: The server-side sample performed slightly worse than the client-side sample in terms of performance, but the differences have clearly narrowed compared to previous tests. The parallel bidding competition proved to be the most effective method.
Testing culture helps find the best solution
Our test results indicated that, in this case, the outcome of the optimal sales was achieved by simultaneously utilising both server-side and client-side implementations.
It is important to bear in mind that the test outcome may not necessarily be applicable to all cases. The performance of different systems varies, as do the audiences, websites, and markets. For this reason, we recommend that publishers develop a testing culture that facilitates easy comparisons between various implementations. This ensures optimal performance and supports the continuous evolution of innovative ad sales.
In addition to the benefits, testing can also be fun and educational, especially when using tools that enable tests to be conducted with low risk, quickly, and easily.
If you would like to learn more about the implementation and testing possibilities of Prebid solutions, please don't hesitate to contact us, and we would be happy to provide further information.
